I'm so happy you're interested in effective climate giving! Here are a few of my favorite resource for learning about environmental/climate giving opportunities:

Giving Green's top climate nonprofits and related research, such as Clean Air Task Force.

Founders Pledge's writing and recommendations around climate philanthropy.

Coefficient Giving's Air Quality and Lead Exposure Action funds and related reports (Air Quality, Lead exposure reduction). 

For a donor starting to explore small gifts to cost-effective climate work, I would recommend the Giving Green Fund.

Feel free to DM me if you have any questions!

EcoResilience Initiative (https://ecoresilienceinitiative.com) works on this, in particular on non-climate-related stuff.  They agree with Giving Green that, counterintuitively, Good Food Institute (usually talked about by animal-welfare fans) might actually be one of the most promising charities in terms of protecting biodiversity and the environment, because if plant-based meat ever took off in a big way it would have massive effects on agricultural land use (ie much less deforestation & habitat destruction would happen).

Hi!

Like you I care about the environment and like you I want to know the most effective way to make the global environment better. 

My focus is on biodiversity, and 80% of the time that means habitat loss. Alternative proteins can halve deforestation permanently, and that's a bigger impact that anything else I know. A full explanation is written up on EcoResilience Initiative's website here.  We think some of the best alternative protein companies for combating future habitat loss are Terra Bioindustries, Hyfe, and Pow.bio Because they are working on changing feedstock for precision fermentation tech away from sugarcane (a tropical crop grown in biodiverse areas) to recycling agricultural waste products like corn husks and spent barley or waste water. Here is EcoResilience Initiative's full write-up on the individual places working directly on the problem. To be clear, the statistic about halving deforestation is still using sugarcane. It could just get even better than that if we recycle agwaste on top of using alternative proteins. Pow.bio is doubling efficiency by changing the fermentation process. Synthesis Capital recommends Hyfe and Pow.bio and some other specific companies too. Incidentally I got up to speed and found a lot of detailed information from GFI, which made me feel like they are pretty good at pushing alternative protein innovation/funding/development forward. Here are all the alternative protein companies. GFI keeps a database. 

If you place more emphasis on longterm technological approaches and solving extinction altogether, biobanking could allow genetic rescue of species suffering extinction debt, and eventually de-extinction. It's also ridiculously cheap. For $3,000,000 for 100 years you could save a species. (For comparison, one study estimates it costs about $1,300,000 per year to keep critically endangered species surviving in the wild with insurance populations in zoos. The authors consider this a low cost.) The Frozen Zoo, Frozen Ark, Svalbard Gloval Seed Vault, and Ocean Genome Legacy are doing this, and Revive and Restore is working on de-extinction tech. 

For a more immediate, less techno-utopia approach, keystone species introductions seem really effective at improving landscapes. I haven't done research into what organizations are doing the best work on this, but the Big Scrub Conservancy is doing some amazing things. They are reintroducing some of the most evolutionarily distinct species and rebuilding an almost lost habitat. Its really exciting. I would particularly search for any freshwater mollusc introduction programs, because those are some highly effective keystone species in some of the most important and most depleted habitats. (And often they are neglected and evolutionarily distinct species themselves too!) Sorry I don't have specific orgs to recommend for this category. My lame excuse is that keystone species introductions don't globalize in quite the same way as the above two biodiversity interventions. 

I might also recommend EDGE for their direct action on neglected species around the globe. They prioritize by evolutionary distinctiveness (unique and irreplaceable species that represent entire branches of the tree of life) which I think is the right approach for a biodiversity crisis.

Giving Green just released their Biodiversity Conservation recommendations and they settled on GFI and Wetlands International. For marine biodiversity they specify "Supporting Implementation and Innovation of Improved Fishing Gear" as one of the most effective ways to reduce overfishing. If you dig in the footnotes you can find examples of people working on this, for example these conservation engineers and this team. I don't think Giving Green wants to claim these are the MOST impactful direct action. That is what their biodiversity philanthropy page is for. There is a lot of uncertainty about management and viability when you drill down this far. But its probably within the top 90% since its within the most effective intervention bracket.

For climate, I think Giving Green (and other EA climate orgs) have that covered. I dug around Giving Green's climate recommendations, trying to find non-policy nonprofits, and I see what you mean. You will probably have to go to the policy orgs Giving Green recommends, and then see if you can dig around and find a specific project/company they endorse. It won't be easy, because I'm guessing these climate policy nonprofits don't want to single out favorites since they will be working with lots of orgs over their lifetime. 

That being said, you might be able to find some specific direct action they are excited about if you search their reports and check their news page. For example, here is a list of 18 geothermal companies put out by Future CleanTech Architects (page 9). First I would pick a sector you expect to be most impactful, and then search within their coverage of that sector. They'll probably highlight a few places acting directly.